Re: CIECAM, ColorSync and WCS
Re: CIECAM, ColorSync and WCS
- Subject: Re: CIECAM, ColorSync and WCS
- From: Chris Cox <email@hidden>
- Date: Mon, 23 Jan 2006 14:07:53 -0800
- Thread-topic: CIECAM, ColorSync and WCS
There seem to be a lot of misunderstandings here.
ColorSync and ICC based color management systems can already use CIECAM02 or
other appearance models - they just encode them in the pre-built profile.
(and several profile makers have been doing so for several years, even with
previous models such as CIECAM97s, LLAB, etc.)
Also, ICC version 4 profiles and CMMs have the option of using any color
appearance model they want to do image specific gamut mapping (but no
shipping CMMs do at this time, partly because it's a harder problem than it
sounds like, partly because everyone is still working on interoperability).
So far, other than XML coding for profiles, I haven't seen anything in WCS
that can't already be accomplished in an ICC version 4 CMM.
Also note that ICC version 4 profiles have measurement data in them (the
device to PCS table) and can be used to construct transforms on the fly -
just like WCS.
You don't have to use CIECAM02 for image specific gamut mapping. It's just
one color appearance model that could be used for gamut mapping. And,
again, that's an option in ICC version 4 CMMs.
Some posters seem to assume that just using CIECAM02 somehow makes WCS
better, or that CIECAM magically solves all the problems. That isn't the
case.
Several studies have shown that CIECAM02 is good for some viewing
conditions, but is not good for mapping between very different viewing
conditions (like the high dynamic range scenes or digital cinema that WCS
claims to support).
CIECAM also does not address simultaneous contrast or other spatial
phenomenon -- it is only addressing isolated colors. Spatial/image
appearance models such as the proposed iCAM are needed to address spatial
phenomenon - and those models are very much in their infancy. And large
dynamic range changes need the spatial/image appearance models.
The publicly proposed architecture for WCS is overkill for 99.99% of users,
more complicated than current ICC based approaches, and based on several
algorithms that have already been proven to have serious problems (for
instance, their default gamut mapping algorithms show serious artifacts and
are judged as having barely acceptable quality). Oh, yeah, and WCS is doing
more work to create the transforms, so it'll most likely be a lot slower.
Someone said that WCS or CIECAM offered "no more cryptic hexadecimal profile
encoding, run-time profile Binding, etc."
Huh?
The current ICC file format is a binary file format -- well documented, easy
to parse, fast to parse (XML ain't that fast) and supported by multiple open
source tools. It is not a human readable format because it doesn't need to
be. And it doesn't need to be ASCII encoded (read: bloated).
And WCS will have more work to do at run time because it has to BUILD the
transform from device profiles before you can use it.
Oh, yeah - some of the details of WCS mean that you could get different
results on different machines when specifying the same devices at either end
of the transform. Prepress folks are going to love that. (I can't add
enough sarcasm here) And having plugin modules for many of the transform
components just means more complexity for the user, or less repeatability
(in addition to slower transforms).
"Canon presented their materials to the ICC first, and the ICC decided not
to adopt it."
That's not what happened. The history would take a long time to explain
(lots of politics, lots of bad behavior). The summary: they presented
something that wasn't what they were supposed to be working on, they based
it on unproven science (some of which has proven NOT to work) and lots of
handwaving, and what they presented required changes incompatible with the
existing frameworks (most of which were not required to achieve the stated
goals).
" It's dispiriting to see how inflexible, parochial and blinkered companies
and standards bodies can prove to be."
Yes, but some companies despise open standards that are not under their
control. And some standards bodies prefer open standards that actually
benefit the end user and are based on solid science and research.
"Based on what I read above, Apple, the ICC, etc., seem unable to learn that
success *for everyone* in color management only comes from agreement and
standardization, and that proprietary systems set us all back."
I'm not sure what you read, but I think you have them backwards.
The ICC is working on open standards that use all the latest color science
knowledge to help users.
WCS is a proprietary "standard".
Overall, I don't see WCS solving any end-user problems, just creating new
problems that users didn't have before.
The ICC approach/implementation has the science. The real remaining
problems are in the workflow and communication (application <-> drivers
being a big problem).
(Hmm, Tom Lianza just posted pretty much the same thing, but with more
detail)
I hope this can help reduce the flood of FUD regarding CIECAM and WCS.
Chris
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden